GNU bug report logs - #73018
31.0.50; wdired + replace-regexp only modifies the visible portion of the buffer

Previous Next

Full log

Message #38 received at 73018 <at> debbugs.gnu.org (full text, mbox):

From: Michael Heerdegen <michael_heerdegen <at> web.de>
To: Juri Linkov <juri <at> linkov.net>
Cc: Madhu <enometh <at> meer.net>, 73018 <at> debbugs.gnu.org
Subject: Re: bug#73018: 31.0.50; wdired + replace-regexp only modifies the
 visible portion of the buffer
Date: Mon, 09 Sep 2024 16:55:59 +0200

[Message part 1 (text/plain, inline)]

Juri Linkov <juri <at> linkov.net> writes:

> M-: (buffer-substring (- (point-max) 2) (- (point-max) 1))
> => #("7" 0 1 (fontified nil invisible dired-hide-details-link))
>
> M-> ;; (end-of-buffer)
>
> M-: (buffer-substring (- (point-max) 2) (- (point-max) 1))
> => #("7" 0 1 (face default dired-symlink-filename t fontified t
> invisible dired-hide-details-link))
>
> And indeed, after going to the end of the Dired buffer
> the last file gets an additional property `dired-symlink-filename'
> used by Isearch/Replace.

So, the properties we use are of different types: some are already
attached by `dired-insert-set-properties', others later by font-lock
(see `dired-font-lock-keywords').  AFAIR this is because some tests are
more expensive than others (e.g. the test whether a link is broken) and
are intentionally delayed.

Would something like this be good?

[0001-WIP-Bug-73018.patch (text/x-diff, inline)]

From c5ef8c8db7d8b2ab03998a0f3f2a1b820ff24e2a Mon Sep 17 00:00:00 2001
From: Michael Heerdegen <michael_heerdegen <at> web.de>
Date: Mon, 9 Sep 2024 16:46:13 +0200
Subject: [PATCH] WIP: Bug#73018

---
 lisp/dired-aux.el | 8 ++++++--
 1 file changed, 6 insertions(+), 2 deletions(-)

diff --git a/lisp/dired-aux.el b/lisp/dired-aux.el
index d79ec342435..77dde7147bc 100644
--- a/lisp/dired-aux.el
+++ b/lisp/dired-aux.el
@@ -3740,8 +3740,12 @@ dired-isearch-search-filenames
 The returned function narrows the search to match the search string
 only as part of a file name enclosed by the text property `dired-filename'.
 It's intended to override the default search function."
-  (isearch-search-fun-in-text-property
-   (funcall orig-fun) '(dired-filename dired-symlink-filename)))
+  (let ((search-fun
+         (isearch-search-fun-in-text-property
+          (funcall orig-fun) '(dired-filename dired-symlink-filename))))
+    (lambda (&rest args)
+      (font-lock-ensure)
+      (apply search-fun args))))

 ;;;###autoload
 (defun dired-isearch-filenames ()
--
2.39.2

[Message part 3 (text/plain, inline)]

A related question is whether everybody always wants to search in
symlink targets when isearching file names in dired... I don't.  Would
it be worth to add an option for that?  Currently the properties are
just hardcoded.

Then, in the above patch we could make the `font-lock-ensure' call
depend on the value of that option.


> Also noticed that doing the first replacement always raises an error:
>
> Debugger entered--Lisp error: (error "Match data clobbered by buffer
> modification hooks")
>   replace-match("!" nil nil)
>   replace-match-maybe-edit("!" nil nil nil (672 673 #<buffer char>) nil)
>   perform-replace("7" "!" t t nil nil nil nil nil nil nil)
>   query-replace-regexp("7" "!" nil nil nil nil nil)
>   funcall-interactively(query-replace-regexp "7" "!" nil nil nil nil nil)
>   command-execute(query-replace-regexp)

Do I interpret the code in replace_match correctly: this error doesn't
even mean the match data has been clobbered - only that modification
hooks called searching functions?  I don't know what the referenced
search_regs.num_regs exactly contains.  But we already seem to ensure
not to clobber match data.


Michael.

Previous Next

GNU bug tracking system
Copyright (C) 1999 Darren O. Benham, 1997,2003 nCipher Corporation Ltd, 1994-97 Ian Jackson.